Financial Barriers to Health Care and Outcomes After Acute Myocardial Infarction FREE

At a time of increasing health care costs, patients commonly face substantial financial barriers to obtaining needed health care. Although much attention has focused on the uninsured, more than 16 million Americans avoid care due to cost or have trouble affording their medications despite having health insurance.¹ This issue is growing in importance as recent budgetary constraints on state and private insurers have added burdens to patients by reducing eligibility, increasing cost-sharing, or limiting benefits such as prescription drug coverage.^2- 6

In the United States, 61% of adults with health insurance currently report difficulty paying their medical bills.⁷ Patients who are challenged by the cost of health care may have an increased risk for adverse health outcomes. Unfortunately, although the negative impact of financial barriers to health care is often stated, few prospective, longitudinal studies have directly investigated this relationship.

Accordingly, we sought to determine if self-reported financial barriers to health care services or medication were associated with worse patient outcomes. We studied recovery after an acute myocardial infarction (AMI), a common medical condition that requires continuing access to health care and guideline-based medications.^8- 9 This study was conducted as part of a multicenter, prospective, observational study of patients' recovery after AMI: the Prospective Registry Evaluating Myocardial Infarction: Event and Recovery (PREMIER).¹⁰ At the time of hospitalization, we asked patients whether they avoided health care services in the prior year due to costs and whether they did not take prescribed medication as instructed due to cost. By directly assessing this information from patients, we focused specifically on patients' self-reported financial burden rather than inferring it from their insurance status or reported income. Patients were followed up over the subsequent year to examine the association of their reported financial barriers with a range of outcomes including mortality, rehospitalization, and health status.

The PREMIER study, as described previously,¹⁰ enrolled 2498 patients with an AMI from 19 medical centers in the United States between January 1, 2003, and June 28, 2004. Patients were aged 18 years or older and had an AMI confirmed by elevated biomarkers of myocardial injury (troponin level or creatine kinase MB fraction) and supporting evidence of an AMI (>20 minutes ischemic signs/symptoms, electrocardiographic ST changes, or both). Eligible patients either presented directly to the enrolling institution or were transferred within the first 24 hours of symptoms. Incarcerated patients and those with elevated cardiac enzyme levels as a complication of elective coronary revascularization were not included.

Patients' sociodemographic, clinical, and treatment data were collected from chart abstractions, baseline interviews administered by trained data collectors within 24 to 72 hours of admission, and follow-up interviews at 12 months. Race/ethnicity was determined by self-report. Institutional research board approval was obtained at each participating institution, and patients signed informed consent for participation in the study.

Patients' health status, obtained by telephone interview, was assessed by their responses to the disease-specific Seattle Angina Questionnaire (SAQ)¹¹ and the generic Short Form-12 (SF-12).¹² The SAQ, a validated instrument, quantifies patients' disease-specific health status through 19 items assessing their symptoms, physical function, and quality of life. The SAQ scores range from 0 to 100 for each category, in which higher scores represent fewer anginal symptoms, better function, and higher quality of life.^11,13- 14 The SF-12 quantifies general functional status by generating both a summary physical component score (PCS) and a mental component score (MCS). For all of these scales, mean differences of more than 5 points are considered clinically significant.^13,15

We also determined all-cause mortality and all-cause and cardiac-specific rehospitalizations. Cardiac-specific rehospitalizations were defined from patient recall of hospitalization for heart failure, AMI, or angina as well as procedures of coronary artery bypass graft surgery or percutaneous coronary intervention. We obtained mortality data by cross-referencing patients' Social Security numbers with the Social Security Death Master File and identified rehospitalizations and health status through follow-up interviews.

To define financial barriers to health care, we asked the following questions: “In the past year, have you avoided obtaining health care services because of cost?” and “In the past year, how often have you not taken a medication that your doctor prescribed because of cost?” Avoiding health care services due to cost was answered either yes or no. Avoidance of medications due to cost was answered on a 5-point Likert scale ranging from “never” to “always.” To create a dichotomous variable, it was categorized into “never” or “rarely” for no vs “occasionally,” “often,” or “always” for yes.

We compared descriptive statistics of demographic and clinical variables using χ² or Fisher exact tests for categorical variables, as appropriate, and t test or analysis of variance for continuous variables. We generated survival estimates using Kaplan-Meier estimates and tested using the log-rank test for patients with financial barriers to health care services and to medications.

To test our hypotheses, we prespecified demographic and clinical variables that we believed should be taken into account in isolating the relationship of financial barriers with outcomes. Through a series of sequential steps we ran a series of models in which we adjusted for each subset of variables to better understand the relationship between these variables and the association between financial barriers and outcomes.

The demographic variables included age, sex, and race. Clinical characteristics included diabetes mellitus; hypertension; tobacco smoking; coronary artery disease (prior AMI, coronary artery bypass graft, or percutaneous coronary intervention); ST-elevation MI vs non–ST-elevation MI; and a prognostic risk score developed as a part of the Cooperative Cardiovascular Project¹⁶ that included cardiac arrest, anterior or lateral location of AMI, systolic blood pressure, white blood cell count, creatinine level, and heart failure.

In the third model for health status outcomes, we adjusted for the patients' baseline health status. For example, in measuring angina frequency at 12 months following an AMI, we adjusted for the patients' baseline angina frequency before their AMI admission. This adjustment produces a statistically equivalent model to one that assesses the change in patients' health status.

We performed a 2-part secondary analysis. In the first part, to test the sensitivity of our method for quantifying financial barriers, we performed an analysis that controlled for the traditional indicators of insurance status, household income, and level of education. Insurance status was dichotomized (yes/no); household income was dichotomized (<$30 000); and level of education was dichotomized (≥grade 12). Due to missing data on income, we also adjusted for missing income as a dummy variable.

In the second part, a separate analysis was performed that adjusted for coronary angiography, coronary revascularization, and the following guideline-based inpatient quality-of-care measures: number of eligible indicators received (aspirin at arrival/discharge, angiotensin-converting enzyme inhibitor for left ventricular systolic dysfunction at discharge, smoking cessation instructions, β-blocker at arrival/discharge),^8- 9 and percentage of eligible indicators received. This permitted us the opportunity to equilibrate the processes of inpatient care and to isolate whether the differences in outcomes are attributable to processes of care that occur after discharge.

For the dichotomous outcome (the presence of angina), we used multivariable hierarchical logistic regression models. To facilitate clinical interpretability, the presence of angina was transformed from the SAQ scores on this domain to a scale of “daily,” “weekly,” “monthly,” or “none.” Since the goal is to be angina free and the presence of angina was skewed with 75% of patients reporting no angina at 12 months, we developed a dichotomous variable as the dependent variable in our multivariable models. For the continuous outcomes of SAQ quality-of-life and SF-12 scores, we used multivariable hierarchical linear regression. For the time-to-event outcomes of rehospitalization and mortality, we used multivariable hierarchical Cox proportional hazards regression or shared frailty models.¹⁷ Hierarchical regression and frailty models accounted for clustering of patients within hospitals. The frailty models included a frailty parameter for each hospital that describes unexplained heterogeneity in survival rates across the hospitals. We tested proportional hazards assumptions for each model and verified them using Schoenfeld residuals.

Patients could be missing follow-up health status data due to death (n = 199 [8.0%]), refusal to participate in the 12-month interview (n = 78 [3.2%]), or loss to follow-up (n = 261 [10.5%]). August 1, 2005, was the date of final follow-up for these analyses. To evaluate the potential bias of those lost to follow-up, we tested for significant differences (P<.05) across various demographic, clinical, and disease characteristics in addition to the previously mentioned variables. We compared the descriptive statistics using χ² or Fisher exact tests for categorical variables, as appropriate, and t test or analysis of variance for continuous variables. After reviewing these variables, we did not identify any further potential confounders with known or suspected associations with post-AMI outcomes. All analyses were performed using SAS version 9.1.3 (SAS Institute Inc, Cary, NC) and R version 2.1.1 (R Foundation for Statistical Computing, Vienna, Austria).

Of the 2498 individuals enrolled in the PREMIER registry, 2439 (97.6%) answered the question for financial barriers to health care services and 2454 (98.2%) answered the question for financial barriers to medication. The prevalence of reported financial barriers to health care services or medication was 18.1% and 12.9%, respectively. Among the 2420 (96.9%) individuals who answered both questions, 9.6% reported financial barriers to both health care services and medication.

There were substantial differences in baseline characteristics between those who reported financial barriers (ie, avoidance due to cost) to health care services or medication and those who did not (Table 1 and Table 2). Individuals most likely to report financial barriers to health care services or medication were more likely to be younger than 65 years of age, female, and nonwhite. Additionally, individuals who reported financial barriers were more likely to have less education, no insurance, and to live with less income than their counterparts. Of the individuals who reported financial barriers to health care services or medication, 68.9% and 68.5%, respectively, had health insurance.

The overall burden of comorbid disease was greater among those with reported financial barriers to health care services or medication. Risk factors such as elevated body mass index, smoking, diabetes mellitus, and past coronary artery disease were all significantly more prevalent among the individuals with reported financial barriers.

Individuals who reported financial barriers to medication had lower rates of coronary revascularization and received a lower percentage of eligible quality-of-care indicators. This was not the case in those with financial barriers to health care services.

At the time patients presented with their AMI, those who reported financial barriers to health care services had a 12.9% higher prevalence of angina before admission (Table 3). They had a 9.4% significantly higher likelihood of severely diminished quality of life with lower mean (SD) SAQ scores and significantly poorer overall health status (SF-12 PCS = 38.6 [13.1] vs 43.4 [12.3] and MCS = 44.3 [12.6] vs 50.7 [11.1]; P<.001 for both).

By 12 months, the prevalence of angina was lower in both groups, but there remained a significant difference between those with and without financial barriers to health care (Table 3). Individuals who reported financial barriers to health care services had a 12.1% higher prevalence of angina (P<.001). Only 73.0% of this group experienced good to excellent quality of life compared with 87.1% of the group that did not report a barrier (P<.001). As a result, the mean SAQ scores and SF-12 scores remained significantly lower for those with financial barriers (P<.001).

Individuals who reported financial barriers had significantly higher unadjusted rehospitalization rates. Reported financial barriers to health care services were associated with an 11.2% higher all-cause rehospitalization rate (P <.001) and an 8.0% higher cardiac rehospitalization rate at 1-year follow-up (P<.001). As demonstrated by the Kaplan-Meier graphs, these differences widened over time (Figure 1).

After adjustment for demographic and clinical factors, poorer health outcomes persisted at 1-year follow-up in those who reported financial barriers to health care services (Figure 2). In the final adjusted models, patients reporting financial barriers were more likely to report persistent angina (odds ratio [OR], 1.33; 95% confidence interval [CI], 0.97-1.83; P = .07), had a 4-point lower SAQ quality-of-life score (mean difference in scores = −4.0; 95% CI, −6.3 to −1.8; P = .001), and had slightly lower SF-12 PCS by 2.2 points (mean difference in scores = −2.2; 95% CI, −3.6 to −0.9; P = .003) and MCS by 2.9 points (mean difference in scores = −2.9; 95% CI, −4.1 to −1.7; P<.001). They also experienced a significantly higher hazard for all-cause rehospitalization (hazard ratio [HR], 1.3; 95% CI, 1.1-1.5; P = .007) and nonsignificantly higher hazard for cardiac rehospitalization (HR, 1.3; 95% CI, 1.0-1.6; P = .06). There was no significant association with mortality (HR, 0.8; 95% CI, 0.5-1.3; P = .35).

In the secondary analysis with additional control for insurance status, level of education, household income, and missing income data, there continued to be significant differences in outcome. Patients reporting financial barriers experienced a 30% higher hazard for all-cause rehospitalization (HR, 1.3; 95% CI, 1.0-1.5; P = .02), and SF-12 MCS remained slightly lower by 2.5 points (mean difference in scores = −2.5; 95% CI, –3.8 to –1.2; P <.001). In the second part of the secondary analysis, when coronary angiography, coronary revascularization, and inpatient quality-of-care measures were taken into account, there were no substantial changes in any of the results.

At the time patients presented with their AMI, individuals who reported financial barriers to medication had a 17.7% higher prevalence of angina before admission (Table 4). They also had an 11.1% significantly higher likelihood of severely diminished quality of life with lower mean (SD) SAQ scores and significantly poorer overall health status (SF-12 PCS = 37.4 [13.0] vs 43.3 [12.3] and MCS = 43.3 [12.8] vs 50.4 [11.3]; P<.001 for both).

In a pattern similar to that of patients who reported financial barriers to health care services, by 12 months the prevalence of angina was lower in both groups, but there remained a significant difference between those with and without financial barriers to medication (Table 4). Individuals who reported financial barriers to medication had a 17% higher prevalence of angina (P<.001). Only 66.2% of individuals with reported financial barriers experienced good to excellent quality of life compared with 86.8% of their counterparts without the barrier (P<.001). As a result, the mean SAQ scores and SF-12 scores remained significantly lower for those with financial barriers (P<.001).

Individuals who reported financial barriers for medication had significantly higher unadjusted rehospitalization rates. Reported financial barriers to medication were associated with a 19.2% higher all-cause rehospitalization rate (P<.001) and a 16.4% higher cardiac rehospitalization rate (P<.001). The Kaplan-Meier graphs demonstrate that these differences widened over time (Figure 1).

After adjustment for potential confounders, poorer health outcomes persisted at 1-year follow-up in those who reported financial barriers to medication (Figure 3). In the final adjusted models, patients reporting financial barriers were more likely to report persistent angina (OR, 1.5; 95% CI, 1.1-2.2; P = .02), had a 7.6-point, clinically significant, lower SAQ quality-of-life score (mean difference in scores = −7.6; 95% CI, −10.2 to −4.9; P<.001), and slightly lower SF-12 PCS by 2.7 points (mean difference in scores = −2.7; 95% CI, −4.3 to −1.1; P = .003) and MCS by 4.4 points (mean difference in scores = −4.4; 95% CI, −5.8 to −3.0; P<.001). They also experienced a higher hazard for all-cause rehospitalization (HR, 1.5; 95% CI, 1.2-1.8; P<.001) and for cardiac rehospitalization (HR, 1.7; 95% CI, 1.3-2.2; P<.001). There was no significant association for mortality (HR, 1.4; 95% CI, 0.9-2.1; P = .10).

In the secondary analysis with additional control for insurance status, level of education, household income, and missing income data there continued to be significant differences in outcome. Patients reporting financial barriers experienced a clinically significant 5.8-point lower SAQ quality-of-life score (mean difference in scores =−5.8; 95% CI, –8.6 to –3.0; P <.001) and slightly lower SF-12 PCS by 1.8 points (mean difference in scores =−1.8; 95% CI, –3.5 to –0.02; P =.05) and MCS by 4.2 points (mean difference in scores = −4.2; 95% CI, –5.7 to –2.7; P<.001). They experienced a higher hazard for all-cause rehospitalization (HR, 1.5; 95% CI, 1.2-1.8; P<.001) and a higher hazard for cardiac rehospitalization (HR, 1.6; 95% CI, 1.2-2.1; P<.001). In the second part of the secondary analysis, when coronary angiography, coronary revascularization, and inpatient quality-of-care measures were taken into account, there were again no substantial changes in any of the results.

Financial barriers to health care, as defined by self-reported avoidance of health care services or medication due to cost, are a common and potent risk factor for adverse outcomes in the AMI population. Almost 1 in 5 patients in our sample reported financial barriers to health care services and 1 in 8 to medication. This patient characteristic was a strong predictor of adverse outcomes, even after controlling for traditional risk factors. Patients with financial barriers had a higher prevalence of angina, worse quality of life, and poorer overall physical and mental function, both at the time of their AMI and 1 year later. The more severe clinical consequences, however, were seen among individuals who reported financial barriers to medication. Patients in this group experienced poorer health status outcomes overall and had a 50% higher hazard for all-cause rehospitalization and a 70% higher hazard for cardiac rehospitalization.

Although we found that many patients had self-reported financial barriers, our results are consistent with estimates from a recent population-based survey from the Kaiser Family Foundation. In that study, 29% of adults, or someone in their household, avoided medical treatment, cut pills, or did not fill a prescription in the past year because of cost.⁷ A separate Kaiser Family Foundation Report demonstrated 1 in 6 women with private health insurance and 1 in 3 with Medicaid reported delays or avoidance of care due to cost.¹⁸ Moreover, 17% of private-, 19% of Medicaid-, and 15% of Medicare-insured women did not fill their prescription due to costs.¹⁸

Given the current climate of increased cost-sharing, the number of underinsured and uninsured individuals will continue to increase,^5,19 and a critical question is whether addressing these financial barriers can improve health outcomes. Although this study cannot prove cause and effect, we postulate that these barriers restrict access to care and adherence to medical regimens. The avoidance of care and medications likely has significant consequences for the AMI population that is dependent on timely access to physicians and sustained treatment with guideline-based medications for secondary prevention and symptom control.^8- 9

The avoidance of medication due to cost was particularly associated with worse outcomes, which is presumably mediated by lack of adherence. Our findings are consistent with several studies that report that nonadherence to medications, secondary to financial barriers, resulted in adverse outcomes.^20- 22 The study by Hsu et al demonstrated that caps on Medicare drug benefits were associated with poorer adherence to drug therapy and resulting higher blood pressure, lipid levels, and glucose levels.²² None of the studies have investigated directly the association of self-reported avoidance of health care or medication due to cost in the AMI population.

The questions used in our study might be understood as a marker for uninsurance while also capturing underinsurance. As a result, we discovered that many patients in this study reported barriers even though they had insurance. In this study, 68.9% and 68.5% of individuals who reported financial barriers to health care services or medication, respectively, were insured; 42.4% and 47.6%, respectively, had Medicaid or Medicare coverage. Thus, although insurance coverage may be important for the population, it may not eliminate financial barriers to care. Our study may be highlighting underinsurance, the definition of which includes “(1) too few services are covered or the coverage is inadequate; (2) amounts of out-of-pocket expenditures, with or without regard to family income, are excessive; (3) insurance is perceived to be inadequate; or (4) some combination is present.”²³ Thus, any plans for health care reform may benefit from an appreciation that many people in the insured population report financial barriers to care.

Since insurance coverage alone does not eliminate financial barriers to health care, we chose to measure self-reported avoidance of health care due to cost. Other analyses of underinsurance avoided direct patient queries and instead used more indirect measures such as the risk of large out-of-pocket expenditures for a catastrophic illness, the proportion of covered claims vs that in the largest federal employee program, or comparing the extent of a patient's out-of-pocket medical costs against his or her income.^24- 25 In addition, numerous studies have used insurance status, household income, or ZIP code^26- 30 as surrogates for financial barriers. This study demonstrates the strength of the direct approach to characterizing these barriers from the patient's perspective.

This strategy of using patient-centered questions regarding barriers to health care is further supported by the findings of our secondary analysis. That analysis, despite additional control for insurance status as well as traditional markers of socioeconomic status including education and household income, continued to uncover significant health disparities. Notably, the overall hazard did not change in all-cause rehospitalization for barriers to health care services or medication. This points to the complementary information that can be gathered on financial barriers to health care when using a direct approach to characterizing these barriers.

Of particular interest, there was concern that the newly uncovered differences in post-AMI outcomes may be due to potential differences in inpatient care between our 2 populations. However, the results of our secondary analysis that controlled for coronary angiography, coronary revascularization, and guideline-based inpatient quality-of-care measures did not substantially change any of our findings. This therefore shifts the focus of disparities in health access to the processes of care pre-AMI and postdischarge.

An equally important contribution of this study is the strong evidence for the consequence of these financial barriers. Our study reveals that these self-reported barriers to care and medication are associated with poorer recovery after an AMI. The findings may be helpful to improve risk stratification of patients and to address structural issues in the health care system predisposing certain patients to worse outcomes. This study provides further support for improved needs assessment and discharge planning combined with a mechanism to facilitate implementation of discharge plans.³¹

There are some issues to consider in the interpretation of this study. Although our objective was to determine the association of baseline self-reported financial barriers with long-term post-AMI outcomes, we did not assess self-reported financial barriers at 12 months. It is therefore possible that the status of some patients either improved or worsened, which might have biased our results to the null. Second, despite being performed across many geographic regions that included both academic and nonacademic institutions, the results of this study still may not be generalized to the entire population in the United States, particularly to rural populations. Third, there exists significant interhospital variability in the type and quality of care delivered by hospitals in the United States and the selected sites in this study may not adequately reflect the broad quality differences in AMI care. Nevertheless, the study did include a diverse set of sites, including both teaching and nonteaching institutions. Fourth, the evaluation of financial barriers relied on self-reporting, which provides information about the patients' perspectives. The responses had strong prognostic importance, but the mechanism by which this is mediated could not be determined with certainty. Last, although we tested for mortality as an outcome measure, the CIs were wide.

In summary, financial barriers to health care are a common and potent risk factor in the AMI population. Focusing specifically on patients' self-reported avoidance of health care services and medications due to cost reveals that these barriers are prominent in individuals with health insurance, suggesting underinsurance. These barriers are associated with, and may contribute to, poorer health status and increased rehospitalization in individuals following an AMI. There is a need to develop approaches that will mitigate this increased risk and address this barrier to care and medications.